How it works / engine flow

How DisEASE works

Every recommendation begins with a verified evidence chain, runs through a structured workup, and is scored by a deterministic rule engine. Dr. EASE explains the result and retrieves the sources behind it. It never changes a score and never adds a recommendation the engine did not produce.

4
disease states live
2,385
scoring rules
991
pharmacist-reviewed passages
0
rankings written by AI
step 01

Evidence chain

Versioned, audited, every change tracked
  1. SourcesGuidelines, openFDA, ClinicalTrials.gov
  2. ChunkedAI-proposed draft
  3. VerifyEdits and approvesPharmacist-in-the-loop
  4. Clinical ConceptsIndexed by meaning
  5. Rule BuilderVersioned and auditable
step 02

Workup

Standardized clinical signals across diseasesNo patient identifiers
  1. Basic InfoAge, Insurance Tier
  2. Vitals & LabsA1c, eGFR, Potassium, 7-Day Fast/PP, 7-Day BP, Volume Status
  3. ComorbiditiesCurrent Conditions
  4. AllergiesDrugs & Contrast
  5. Current TherapiesActive Medication Regimen
step 03

Parallel engines

Deterministic score and generative explanation, in parallel

Track A / deterministic

Patent pending

Scoring Engine

No generative AI in scoring. Same inputs, same ranked list, every time.

  • Safety screening
  • Coverage signals
  • Transparent trace log
  • Clinical alerts

Track B / generative AI

Dr. EASE

Cites its sources. Never assigns scores.

OrchestratorSpecialists, dispatched per question

The orchestrator dispatches only the specialists a question needs. They run in parallel, and their output is merged into one cited answer.

0rankings written by AI

system map

The same engine, standing still

The three steps above are a sequence: what happens to a case. These are the same parts as layers: what a clinician touches, what runs underneath it, and what both tracks read from.

Applications

What the clinician touches

  • Ask Dr. EASECurbside consult
  • WorkupDetailed in step 02
  • RecommendationsRanked, with a trace
  • Dynamic Clinical NotesSOAP note drafted from the case

Platform

Detailed in step 03

  • Scoring EngineDeterministic
  • Dr. EASEGenerative

Data

Produced by step 01

Clinical KnowledgeVersioned and auditablePharmacist-in-the-loop
2,385
Scoring rules across 4 disease states
991
Pharmacist-reviewed passages
  • Standard Clinical Model
  • Clinical weights
  • Evidence chunks
  • Memory entries

What a trace looks like

Every score is the sum of the rules that fired, and the engine reports each one by its deployed id. These rows are real: they are the rules that acted on the winning option in a published case.

trace / Hypertensionreproducible
boost-Chlorthalidone-1Dose optimization of active therapy to reduce polypharmacy
boost-Chlorthalidone-2Most potent first-line agents for severe uncontrolled hypertension
penalty-Chlorthalidone-3Thiazides increase uric acid; may trigger gout flares (lowered the score)
Rules that fired on increase Chlorthalidone (Thalitone) in a published proof case. Rules that were evaluated and did not fire are not part of the published output, so none are shown here. The full trace, with its citations, is on the walkthrough.

Walk through the whole case

See it on a real case

Twelve complete workups, every rule that fired, and the source behind each one. No sign-in.

Explore a scored caseCreate an account